import math
import arvados_cwl.util
-import ruamel.yaml as yaml
+import ruamel.yaml
from cwltool.errors import WorkflowException
from cwltool.process import UnsupportedRequirement, shortname
logger = logging.getLogger('arvados.cwl-runner')
metrics = logging.getLogger('arvados.cwl-runner.metrics')
+def cleanup_name_for_collection(name):
+ return name.replace("/", " ")
+
class ArvadosContainer(JobBase):
"""Submit and manage a Crunch container request for executing a CWL CommandLineTool."""
container_request["state"] = "Committed"
container_request.setdefault("properties", {})
+ container_request["properties"]["cwl_input"] = self.joborder
+
runtime_constraints = {}
if runtimeContext.project_uuid:
container_request["container_image"] = arv_docker_get_image(self.arvrunner.api,
docker_req,
runtimeContext.pull_image,
- runtimeContext.project_uuid,
- runtimeContext.force_docker_pull,
- runtimeContext.tmp_outdir_prefix,
- runtimeContext.match_local_docker,
- runtimeContext.copy_deps)
+ runtimeContext)
network_req, _ = self.get_requirement("NetworkAccess")
if network_req:
if api_req:
runtime_constraints["API"] = True
+ use_disk_cache = (self.arvrunner.api.config()["Containers"].get("DefaultKeepCacheRAM", 0) == 0)
+
+ keep_cache_type_req, _ = self.get_requirement("http://arvados.org/cwl#KeepCacheTypeRequirement")
+ if keep_cache_type_req:
+ if "keepCacheType" in keep_cache_type_req:
+ if keep_cache_type_req["keepCacheType"] == "ram_cache":
+ use_disk_cache = False
+
runtime_req, _ = self.get_requirement("http://arvados.org/cwl#RuntimeConstraints")
if runtime_req:
if "keep_cache" in runtime_req:
- runtime_constraints["keep_cache_ram"] = math.ceil(runtime_req["keep_cache"] * 2**20)
+ if use_disk_cache:
+ # If DefaultKeepCacheRAM is zero it means we should use disk cache.
+ runtime_constraints["keep_cache_disk"] = math.ceil(runtime_req["keep_cache"] * 2**20)
+ else:
+ runtime_constraints["keep_cache_ram"] = math.ceil(runtime_req["keep_cache"] * 2**20)
if "outputDirType" in runtime_req:
if runtime_req["outputDirType"] == "local_output_dir":
# Currently the default behavior.
if runtimeContext.submit_runner_cluster:
extra_submit_params["cluster_id"] = runtimeContext.submit_runner_cluster
- container_request["output_name"] = "Output for step %s" % (self.name)
+ container_request["output_name"] = cleanup_name_for_collection("Output from step %s" % (self.name))
container_request["output_ttl"] = self.output_ttl
container_request["mounts"] = mounts
container_request["secret_mounts"] = secret_mounts
for pr in output_properties_req["outputProperties"]:
container_request["output_properties"][pr["propertyName"]] = self.builder.do_eval(pr["propertyValue"])
else:
- logger.warning("%s API server is too old to support setting properties on output collections.",
- self.arvrunner.label(self))
+ logger.warning("%s API revision is %s, revision %s is required to support setting properties on output collections.",
+ self.arvrunner.label(self), self.arvrunner.api._rootDesc["revision"], "20220510")
if runtimeContext.runnerjob.startswith("arvwf:"):
wfuuid = runtimeContext.runnerjob[6:runtimeContext.runnerjob.index("#")]
if container["output"]:
outputs = done.done_outputs(self, container, "/tmp", self.outdir, "/keep")
+
+ properties = record["properties"].copy()
+ properties["cwl_output"] = outputs
+ self.arvrunner.api.container_requests().update(
+ uuid=self.uuid,
+ body={"container_request": {"properties": properties}}
+ ).execute(num_retries=self.arvrunner.num_retries)
except WorkflowException as e:
# Only include a stack trace if in debug mode.
# A stack trace may obfuscate more useful output about the workflow.
class RunnerContainer(Runner):
"""Submit and manage a container that runs arvados-cwl-runner."""
- def arvados_job_spec(self, runtimeContext):
+ def arvados_job_spec(self, runtimeContext, git_info):
"""Create an Arvados container request for this workflow.
The returned dict can be used to create a container passed as
"kind": "collection",
"portable_data_hash": "%s" % workflowcollection
}
+ elif self.embedded_tool.tool.get("id", "").startswith("arvwf:"):
+ uuid, frg = urllib.parse.urldefrag(self.embedded_tool.tool["id"])
+ workflowpath = "/var/lib/cwl/workflow.json#" + frg
+ packedtxt = self.loadingContext.loader.fetch_text(uuid)
+ yaml = ruamel.yaml.YAML(typ='safe', pure=True)
+ packed = yaml.load(packedtxt)
+ container_req["mounts"]["/var/lib/cwl/workflow.json"] = {
+ "kind": "json",
+ "content": packed
+ }
+ container_req["properties"]["template_uuid"] = self.embedded_tool.tool["id"][6:33]
else:
- packed = packed_workflow(self.arvrunner, self.embedded_tool, self.merged_map, runtimeContext)
+ main = self.loadingContext.loader.idx["_:main"]
+ #print(json.dumps(main, indent=2))
workflowpath = "/var/lib/cwl/workflow.json#main"
container_req["mounts"]["/var/lib/cwl/workflow.json"] = {
"kind": "json",
- "content": packed
+ "content": main
}
- if self.embedded_tool.tool.get("id", "").startswith("arvwf:"):
- container_req["properties"]["template_uuid"] = self.embedded_tool.tool["id"][6:33]
+
+ container_req["properties"].update({k.replace("http://arvados.org/cwl#", "arv:"): v for k, v in git_info.items()})
properties_req, _ = self.embedded_tool.get_requirement("http://arvados.org/cwl#ProcessProperties")
if properties_req:
if runtimeContext.enable_preemptible is False:
command.append("--disable-preemptible")
+ if runtimeContext.varying_url_params:
+ command.append("--varying-url-params="+runtimeContext.varying_url_params)
+
+ if runtimeContext.prefer_cached_downloads:
+ command.append("--prefer-cached-downloads")
+
+ if self.fast_parser:
+ command.append("--fast-parser")
+
command.extend([workflowpath, "/var/lib/cwl/cwl.input.json"])
container_req["command"] = command
def run(self, runtimeContext):
runtimeContext.keepprefix = "keep:"
- job_spec = self.arvados_job_spec(runtimeContext)
+ job_spec = self.arvados_job_spec(runtimeContext, self.git_info)
if runtimeContext.project_uuid:
job_spec["owner_uuid"] = runtimeContext.project_uuid
projects / arvados.git / blobdiff